Pancreatic cancer is the fifth most common cause of cancer in the U.S. The majority of patients present with incurable disease, and of those that are candidates for surgical resection by pancreaticoduo-denectomy, few survive five years. In colorectal cancer also, earlier diagnosis would be of great value. The board objective of this proposal is to explore the possibility of practical screening and diagnostic tests for pancreatic and colorectal cancer based on detection of mutant genes in routine clinical samples such as stool, so as to allow earlier diagnosis. Specifically, this study will detect ras gene mutations in such samples, compare results in pancreatic cancer and in colorectal neoplasms, determine whether atypical duct lesions are the precursor of pancreatic cancer and whether they are the source of the cells which enter the gastrointestinal tract, and develop second stage stool assays based on other mutant genes to serve as tests for confirmation of the ras assay results. These studies will include patients with colorectal adenomas and carcinomas, resectable and unresectable pancreatic exocrine adenocarcinoma, benign pancreatic duct disease, chronic pancreatitis, and colorectal adenomas and carcinomas. Three forms of assay are available, and positive results in one will be confirmed using a different system. This study will focus on routine samples such as stool and blood, and will not emphasize invasive and expensive diagnostic samples such as pancreatic secretions and cytologic samples. The methods used to sensitively detect rare mutant DNA sequences in routine specimens will be readily transferable to the more specialized clinical material. The aim is to develop clinical tests for general use that will allow a real impact on early diagnosis. Aspects of this study benefit directly from the findings of the companion project 2.